Aim to mix assets, don’t match them

What goes up must come down. This is a law of mathematics, statistics and even physics.

Applied to a set of numbers, the word “mean” is a fancy term for “average.” “Reversion” means “a reversal or return.” So mean reversion simply means that through time, something goes back to its average. As applied to investments, the concept of mean reversion or reversion to the mean says that over a period of several investment cycles, most assets’ returns will tend to generate their long-term average returns.

For example, when Japanese equities’ returns in the mid- to late-1980s were far above their long-term average, at some point, it became increasingly likely that they would revert to their mean returns by producing returns that were near, or even far below, their long-term average. This is just what happened from 1989 until early 2003. In the 1990s, the U.S. stock market, particularly technology stocks, returned far more than average.

For the first part of the new century, they reverted to the mean and returned far less than the long-term average. High-quality bonds underperformed in the last years of the 1990s but then became a high-returning investment in the 2000-04 period.

Almost every time an asset class goes too far in one direction, it eventually wings the other way — the only questions are: how far, and for how long? Among the most difficult things to know about mean reversion are how long it will take for above-average returns to fall back to or below their long-term averages, or for below-average returns to increase to or above their long-term averages.

Financial history confirms the difficulty of determining the persistence of returns versus the reversal of returns. Just because an asset class has been outperforming the average does not mean there is some fixed timetable for prices to drop and then recover or the other way around. Just ask anyone who was short tech stocks during their meteoric rise or long Japanese stocks during their prolonged downturn.

Keep in mind that when the reversing move does occur, prices will often push far past their long-term average in the opposite direction. In Wall Street parlance, “reversion to the moon” means that returns keep on surpassing to the upside, while “reversion to the moan” is used when talking about returns staying lower far longer than expected.

Let’s take a quick look at how reversion to the mean can help you keep your portfolio in line with your goals and, more importantly, keep you out of trouble. One of the key measures I use to track equities is the proportion of an individual industry group as a percentage of the Standard & Poor’s 500 stock index.

This is a fairly simple tool that gives me an indication when a group seems to have grown too far or have shrunk too much. Take technology stocks in the 1990s. In 1995, tech stocks composed just about 10% of the S&P 500. By 1999, tech stocks had grown to a whopping 28% before topping out at well over 30% in early 2000.

Without making any judgments about the individual stocks, one could see that the group as a whole had grown far beyond its average weighting and was more than ready to revert back to the mean. If you had followed this measure and decided to cut back your allocation to the group, you would have saved yourself considerable pain. At the same time, energy had fallen from its average 9% to 10% level down to just 5%.

An investor paying attention to mean reversion would have considered switching from tech to energy stocks. With no further economic or financial analysis, just being aware of this concept would have led you to do exactly the right thing for your portfolio.

As another example of mean reversion protecting you, by late 2006 financial stocks had grown to 21% of the S&P 500, up from below 10% several years earlier. Investors who noted this might have considered cutting back financials in their portfolio. As we now know, that would have been exactly the right thing to do based on what happened to financial stocks in 2007 and then into 2008.

“Risk” is one of those four-letter words that has more definitions than grammatical uses. There are many types of risk, so what are we talking about, exactly, when we discuss risk?

Investing, not unlike life, presents each of us with a wide range of risks. There is purchasing power risk, which is the loss of value due to inflation. There is price risk. Prices change daily for many assets, and not always in the direction we would like. Interest rate risk represents the impact of interest rate changes on our fixed-income assets.

Credit risk is the risk we take that a company in which we invest cannot repay what it owes us. Assets can be revalued or devalued, currencies fluctuate against each other, markets rise and fall, and the economy is subject to fluctuation and change. We need always to be aware of what risks we are facing and what risks we are taking.

In investing, there are two basic ways to approach risk. One is to be risk-seeking and select investments that have higher historic volatility in the hopes of earning higher returns. The other is to be risk-averse, completely avoiding investments that have any chance of loss.

Each of these postures has a significant impact on portfolio returns. The old adage “no risk, no return” is truest of all in the financial markets.

Returns are what it is all about. The whole point of asset allocation and investing is to protect your capital and earn a satisfactory rate of return. Your return may come from dividends and interest or it could be a result of long-term gains in asset values, such as the appreciation of stocks or real estate.

Some returns from asset classes are fixed in amount, while others are less predictable and are market-dependent.

Estimating future returns for asset classes is a blend of art and science, past and present. Most asset class return projections start by taking into consideration the actual total returns generated over the most recent one-, three-, five-, 10- and 20-year time periods. There are different ways of projecting returns for different asset classes.

For equity asset classes, projected returns are based on forecasts of real earnings growth, the expected inflation rate, changes in the price-earnings ratio and the dividends anticipated over a given holding period. For fixed-income asset classes, the variables used to calculate returns include current interest rates and inflation rates, the credit strength of the issuer and expected changes in interest rates and inflation.

One of the most frequently used concepts in the process of asset allocation is the notion of standard deviation.

In driving terms, standard deviation is a relatively straightforward way of measuring how far your car tends to steer to the right and to the left of the centerline of the highway.

In financial terms, we calculate the average return of an asset over a specified time period. Then we measure how far away and how often the returns have been above or below the average. This calculation will give us an idea of how volatile an asset is and how far away from the average it has swung. Often this can help us determine when a specific asset class might be about to revert to its mean.

In statistical terms, we know something pretty important about figuring standard deviation.

Let’s take the case of real estate investment trusts at the end of 2006. The index of REIT shares had an average 10-year return of 14%, with a standard deviation of 17%. We know from statistics that 68% of the time, you can expect that REITs will range in returns from 31% (14% + 17%) to -3% (14% – 17%) a year.

When we look at the 2006 return of 31%, we see that we were well outside the range of normal returns (14%), and thus we probably want to look at lightening our portfolio’s exposure to this group. On rare occasions, we may experience an annual return of two standard deviations away from the mean or average return. This happens only about 5% of the time, according to statistical probabilities.

So if, for example, we look at the S&P 500 and see that the 10-year average return is 8.5%, with a standard deviation of 19.1%, we should expect returns to range from -10.6% to 27.6% about 95% of the time. If we see in the newspapers that the stock market is down 25% for the year, we know that we are near one of these rare two standard-deviation events, and we would probably want to be a buyer of the market. This tool works on less volatile assets as well.

At the end of 2006, Treasury inflation protected securities had an annual return of 6.2%, with a standard deviation of 4.5%. Looking at the 2006 calendar year return of just 0.4% gave you a pretty good indication that TIPS returns might be about to rise, and you should consider shifting some of your portfolio’s assets into that area.

You do not have to know how to calculate the standard deviation of an asset class; you just have to know how to use it. Your financial advisers or your other sources of financial advice should be able to provide you with historical standard-deviation numbers for various asset classes.

In an asset allocation and investing context, we often look at the relationship between assets. We want to know if a specific asset behaves the same way as another in response to the same economic conditions. People care greatly about correlations because the benefits of asset allocation depend to a significant degree on our ability to find asset classes with correlations of returns that are stable and low (or negative, if possible).

David M. Darst is managing director and chief investment officer for the global-wealth-management group at Morgan Stanley of New York.